3D-2D silhouette-based image registration for comparative radiography-based forensic identification
作者：
Highlights：
• 2D radiographs and 3D scans of any bone or cavity are automatically superimposed.
• The method considers up to 7 degrees of freedom and occlusions.
• The method has been validated with synthetic and real data.
• The mean overlapping error on synthetic cases is lower than the 2% of the pixels.
• Real frontal sinuses cases are correctly identified in the 58% of the experiments.
摘要
•2D radiographs and 3D scans of any bone or cavity are automatically superimposed.•The method considers up to 7 degrees of freedom and occlusions.•The method has been validated with synthetic and real data.•The mean overlapping error on synthetic cases is lower than the 2% of the pixels.•Real frontal sinuses cases are correctly identified in the 58% of the experiments.
论文关键词：Forensic identification,Comparative radiography,3D-2D image registration,Computer vision,Evolutionary computation
论文评审过程：Received 16 April 2018, Accepted 15 June 2018, Available online 18 June 2018, Version of Record 28 June 2018.
论文官网地址：https://doi.org/10.1016/j.patcog.2018.06.011
